Time required to write down "wisdom" (Re: "Adult supervision")

Thomas Narten narten at us.ibm.com
Tue May 13 11:53:08 CEST 2003


Hi John.

> I think my original mail sounded a bit more cranky than I wanted it,
> but I was more responding to Harald's original mail - by the
> way is there an archive of this list?

Email can be funny that way...

> Harald mentioned that the document he was talking about sat around,
> the token passed between a number of editors, etc.  It sounded to
> me like the process for editing the document, soliciting comments,
> producing revisions and so forth, fell apart, resulting in a
> much longer than needed process.

Yep. 

> I am not advocating pushing documents forward without adequate
> review, but much more for having pro-active editors, WG chairs
> and sheparding ADs to make sure the documents are getting done
> in a timely fashion, especially ones identified as important.
> This is more about breakdown in management, in my opinion.

Yep. I think it would be good to understand when and why this process
seems to breakdown, as it leads to stalled documents much more
frequently than we perhaps realize. And these breakdowns can lead to
an extra month or two or three of delay. Or worse.

A couple more things I might add to me original post.

Thomas Narten <narten at us.ibm.com> writes:

> > It seems that the 'process' has some how gotten twisted to disable the
> > quick publication of important documents.

> This may true if the assumption is that the *process* is what caused
> the problems.

> >From my perspective, the real issue often tends to be:

> 1) Good documents don't pop out in the -00 version.

> 2) Iteration is essential. Iteration means a small number of people
>    (e.g., 1-5) read the document, provide good feedback, and then a
>    new revision is produced.

One thing that needs to be done here as well is that the original
person who reviewed the document needs to sign off on the
revision. That is one of the essential parts of our process (when it
works). Person reviews, provides feedback, a new revision comes a
long, and original reviewer says "works for me". Everyone is happy,
document is truely more of a consensus document. 

But, this doesn't always happen. What sometimes happens is the
revision is so long in coming, it becomes hard to get the attention of
the original reviewer, who has since forgotten about the details of
the document (in the worst case, this leads to the death spiral where
the delays get increasing longer...). Or, the author revises the
document, and *thinks* they have addressed the comments, but there is
no confirmation. When the original reviewer goes back, they may not be
happy with the result. Now there is bad feeling on top of everything
else, in that a reviewer feels like their input was ignored and not
given proper consideration.

Or, the author just responds to the issues clarifying them in
email. While this might explain what is meant, it doesn't mean the
document has gotten any better. Author thinks input has been dealt
with "I responded to that", while the reviewer feels like they have
been ignored. Document further stalls while it is unclear who has the
token...

I'm sure there are other examples. I think your point is right though
that there is a document "management" function that we often don't pay
sufficient attention to, and this does lead to long delayed documents.

Oh, and note also that one also needs to deal with the situation where
the author/reviewer can't agree, so that we don't get deadlock. This
also can block documents for inappropriately long periods of time.

> 3) process is repeated at least a few more times, with a different set
>    of reviewers providing the review and feedback each time.

> 4) Process terminates, because subsequent reviews don't uncover
>    significant issues and the reviewers think the document is good
>    enough to ship.

> You can't rush a document (if you want it to be good). Indeed, when I
> write documents, I personally find that if I reread something I wrote
> a month earlier, I often find obvious things that need fixing. I often
> don't see these if I review the document a few days after last working
> on it. The point here is that good documents just don't happen on the
> first version and time is needed to properly review and iterate.

> Where the "process" sometimes goes wrong is that the sequence of
> reviews and iterations haven't happened properly/optimally. Either not
> enough iterations, or too long between iterations.

> IMO, there is a problem here that bears further examination. Getting
> good reviews and then subsequent revisions in a timely fashion is
> something I see too much of.

> > I think that we, as an organization, do need to do better.

> Yep. IMO, we should look hard at ways of ensuring that the needed
> iteration on revisions happens in a timely fashion. But not too
> timely, as that leads to documents being pushed forward before they
> are truly ready.

> Thomas


More information about the Problem-statement mailing list